Hugging Face Daily Papers·· 27 天前AI 评分58
TraceTrial-Bench:AI Agent Harness 可靠性与故障归因评测基准
Trial by Trace: Benchmarking AI Agent Harnesses for Reliability and Failure Attribution
AI 导读
针对现有基准多关注模型能力而忽视 Agent Harness 评估的问题,该研究提出 TraceTrial-Bench,从上下文、执行、治理等七个维度构建 95 个专家级任务。
来源:Hugging Face Daily Papers · arxiv.org